firefly-linux-kernel-4.4.55.git
11 years agoipack/devices/ipoctal: add missing rx_enable = 1
Samuel Iglesias Gonsalvez [Fri, 18 Jan 2013 07:57:21 +0000 (08:57 +0100)]
ipack/devices/ipoctal: add missing rx_enable = 1

There was a bug in the code when managing a GE IP-OCTAL-485 board. The RX would
be enabled but we have a wrong state in the rx_enable flag.

Then, if the user changes the terminal settings, RX would not be enabled again.

Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agotools: hv: Use CLOEXEC when opening kvp_pool files
Tomas Hozza [Fri, 18 Jan 2013 14:23:41 +0000 (15:23 +0100)]
tools: hv: Use CLOEXEC when opening kvp_pool files

Use CLOEXEC flag when opening kvp_pool_x files to prevent file
descriptor leakage. Not using it was causing a problem when
SELinux was enabled.

Signed-off-by: Tomas Hozza <thozza@redhat.com>
Acked-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Bind all vmbbus interrupts to the boot CPU
K. Y. Srinivasan [Fri, 18 Jan 2013 05:00:51 +0000 (21:00 -0800)]
Drivers: hv: Bind all vmbbus interrupts to the boot CPU

The default interrupt delivery model in Linux does not support the Hyper-V
vmbus delivery model when the guest is configured with multiple VCPUs. I have
sent a patch to address this - delivering the vmbus interrupt on a separate
IDT vector. Until this patch is applied, bind all vmbus interrupts to the boot
CPU.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: vmbus_flow_handler() can be static
Fengguang Wu [Fri, 18 Jan 2013 01:09:32 +0000 (09:09 +0800)]
Drivers: hv: vmbus_flow_handler() can be static

Signed-off-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoMerge 3.8-rc4 into char-misc-next
Greg Kroah-Hartman [Fri, 18 Jan 2013 17:31:04 +0000 (09:31 -0800)]
Merge 3.8-rc4 into char-misc-next

This brings in all of the mei and other fixes that are needed to continue
development in this branch.

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoLinux 3.8-rc4
Linus Torvalds [Fri, 18 Jan 2013 03:25:45 +0000 (19:25 -0800)]
Linux 3.8-rc4

11 years agontb: remove __dev* markings
Greg Kroah-Hartman [Fri, 18 Jan 2013 03:17:42 +0000 (19:17 -0800)]
ntb: remove __dev* markings

These are now gone from the kernel, so remove them from the newly-added
drivers before they start to cause build errors for people.

Cc: Jon Mason <jon.mason@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agonet: Add support for NTB virtual ethernet device
Jon Mason [Sat, 17 Nov 2012 02:27:13 +0000 (19:27 -0700)]
net: Add support for NTB virtual ethernet device

A virtual ethernet device that uses the NTB transport API to
send/receive data.

Signed-off-by: Jon Mason <jon.mason@intel.com>
Reviewed-by: Nicholas Bellinger <nab@linux-iscsi.org>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoPCI-Express Non-Transparent Bridge Support
Jon Mason [Sat, 17 Nov 2012 02:27:12 +0000 (19:27 -0700)]
PCI-Express Non-Transparent Bridge Support

A PCI-Express non-transparent bridge (NTB) is a point-to-point PCIe bus
connecting 2 systems, providing electrical isolation between the two subsystems.
A non-transparent bridge is functionally similar to a transparent bridge except
that both sides of the bridge have their own independent address domains.  The
host on one side of the bridge will not have the visibility of the complete
memory or I/O space on the other side of the bridge.  To communicate across the
non-transparent bridge, each NTB endpoint has one (or more) apertures exposed to
the local system.  Writes to these apertures are mirrored to memory on the
remote system.  Communications can also occur through the use of doorbell
registers that initiate interrupts to the alternate domain, and scratch-pad
registers accessible from both sides.

The NTB device driver is needed to configure these memory windows, doorbell, and
scratch-pad registers as well as use them in such a way as they can be turned
into a viable communication channel to the remote system.  ntb_hw.[ch]
determines the usage model (NTB to NTB or NTB to Root Port) and abstracts away
the underlying hardware to provide access and a common interface to the doorbell
registers, scratch pads, and memory windows.  These hardware interfaces are
exported so that other, non-mainlined kernel drivers can access these.
ntb_transport.[ch] also uses the exported interfaces in ntb_hw.[ch] to setup a
communication channel(s) and provide a reliable way of transferring data from
one side to the other, which it then exports so that "client" drivers can access
them.  These client drivers are used to provide a standard kernel interface
(i.e., Ethernet device) to NTB, such that Linux can transfer data from one
system to the other in a standard way.

Signed-off-by: Jon Mason <jon.mason@intel.com>
Reviewed-by: Nicholas Bellinger <nab@linux-iscsi.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoVMCI: include slab.h into files using kmalloc/kfree
Dmitry Torokhov [Thu, 10 Jan 2013 23:41:43 +0000 (15:41 -0800)]
VMCI: include slab.h into files using kmalloc/kfree

Do not rely on implicit header dependencies as they are known to
break.

Reported-by: Randy Dunlap <rdunlap@infradead.org>
Acked-by: Andy King <acking@vmware.com>
Signed-off-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoVMCI: rename PPNset to ppn_set to avoid camel case
Dmitry Torokhov [Thu, 10 Jan 2013 23:41:42 +0000 (15:41 -0800)]
VMCI: rename PPNset to ppn_set to avoid camel case

Acked-by: Andy King <acking@vmware.com>
Signed-off-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoVMCI: Fix "always true condition"
Andy King [Thu, 10 Jan 2013 23:41:41 +0000 (15:41 -0800)]
VMCI: Fix "always true condition"

vmci_send_datagram() returns an int, with negative values indicating failure.
But we store it locally in a u32, which makes comparison of >= 0 useless.
Fixed to use an int.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Andy King <acking@vmware.com>
Signed-off-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoVMCI: Fix deref before NULL-check of queuepair ptr
Andy King [Thu, 10 Jan 2013 23:41:40 +0000 (15:41 -0800)]
VMCI: Fix deref before NULL-check of queuepair ptr

Check for a valid queuepair ptr before trying to lock the queuepair (which will
deref it).

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Andy King <acking@vmware.com>
Signed-off-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoVMCI: Remove dependency on BLOCK I/O
Andy King [Thu, 10 Jan 2013 23:41:39 +0000 (15:41 -0800)]
VMCI: Remove dependency on BLOCK I/O

No need to bring in dm-mapper.h and along with it a dependency on BLOCK I/O
just to use dm_div_up().  Just use the existing DIV_ROUND_UP().

Reported-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andy King <acking@vmware.com>
Signed-off-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoVMCI: Add PCI as a dependency
Andy King [Thu, 10 Jan 2013 23:41:38 +0000 (15:41 -0800)]
VMCI: Add PCI as a dependency

Add PCI as a dependency to our build, since we always compile in the guest-side
PCI device support.

Reported-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Andy King <acking@vmware.com>
Signed-off-by: Dmitry Torokhov <dtor@vmware.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agopcmcia/vrc4171: Add missing spinlock init
Jean Delvare [Sun, 16 Dec 2012 21:00:50 +0000 (22:00 +0100)]
pcmcia/vrc4171: Add missing spinlock init

It doesn't seem this spinlock was properly initialized. This bug was
introduced by commit 7a410e8d4d97457c8c381e2de9cdc7bd3306badc.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agopcmcia: i82092: fix i82092aa_pci_remove()
Dan Carpenter [Fri, 14 Dec 2012 15:01:08 +0000 (18:01 +0300)]
pcmcia: i82092: fix i82092aa_pci_remove()

Smatch complains because the call to
pci_set_drvdata(dev, &sockets[i].socket);
is reading one step beyond the end of the sockets[] array.  It will
crash when we use it later.

The only place which uses pci_get_drvdata() is i82092aa_pci_remove().
That function should loop through all the sockets and unregister them.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agomei: Fix some more kernel-doc typos in hw-me.c
Sedat Dilek [Thu, 17 Jan 2013 18:54:15 +0000 (19:54 +0100)]
mei: Fix some more kernel-doc typos in hw-me.c

Signed-off-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agotools: hv: Fix how ifcfg-* file is created
Tomas Hozza [Sun, 13 Jan 2013 21:27:40 +0000 (22:27 +0100)]
tools: hv: Fix how ifcfg-* file is created

Fix for the daemon code and for hv_set_ifconfig.sh script, so
that the created ifcfg-* file is consistent with initscripts
documentation.

Signed-off-by: Tomas Hozza <thozza@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agotools: hv: fix a typo in hv_set_ifconfig.sh
Jason Wang [Sat, 5 Jan 2013 05:03:06 +0000 (13:03 +0800)]
tools: hv: fix a typo in hv_set_ifconfig.sh

Signed-off-by: Jason Wang <jasowang@redhat.com>
Acked-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: remove unused variable in vmbus_recvpacket_raw()
Wei Yongjun [Tue, 4 Dec 2012 05:15:42 +0000 (00:15 -0500)]
Drivers: hv: remove unused variable in vmbus_recvpacket_raw()

The variable userlen is initialized but never used
otherwise, so remove the unused variable.

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Cleanup and consolidate reporting of build/version info
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:59 +0000 (06:46 -0800)]
Drivers: hv: Cleanup and consolidate reporting of build/version info

Now, cleanup and consolidate reporting of host and vmbus version numbers.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Capture the host build information
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:58 +0000 (06:46 -0800)]
Drivers: hv: Capture the host build information

Capture the host build information so it can be presented along with the
negotiated vmbus version information.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Implement flow management on the send side
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:57 +0000 (06:46 -0800)]
Drivers: hv: Implement flow management on the send side

Implement flow management on the send side. When the sender is blocked, the reader
can potentially signal the sender to indicate there is now room to send.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Enable protocol negotiation with win8 hosts
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:56 +0000 (06:46 -0800)]
Drivers: hv: Enable protocol negotiation with win8 hosts

Now that we have implemented all of the Win8 (WS2012) functionality, negotiate
Win8 protocol with the host.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Add a check to deal with spurious interrupts
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:55 +0000 (06:46 -0800)]
Drivers: hv: Add a check to deal with spurious interrupts

We establish the handler before we have fully initialized the VMBUS state.
Deal with spurious interrupts.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Handle vmbus interrupts concurrently on all cpus
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:54 +0000 (06:46 -0800)]
Drivers: hv: Handle vmbus interrupts concurrently on all cpus

Vmbus interrupts are unique in that while the interrupt is delivered on a
given vector, these can be handled concurrently on different CPUs. Handle the
vmbus interrupts concurrently on all the CPUs.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Manage event tasklets on per-cpu basis
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:53 +0000 (06:46 -0800)]
Drivers: hv: Manage event tasklets on per-cpu basis

Now that we can potentially take vmbus interrupts on any CPU, make the
tasklets per-CPU.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Get rid of unnecessary request for offers
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:52 +0000 (06:46 -0800)]
Drivers: hv: Get rid of unnecessary request for offers

This call to seek offers is not necessary and just adds unnecessary delay.
Get rid of it.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Get rid of the unused global signaling state
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:51 +0000 (06:46 -0800)]
Drivers: hv: Get rid of the unused global signaling state

Now that we have implemented a per-connection signaling mechanism, get rid
of the global signaling state. For hosts that don't support per-connection
signaling handle, we have moved the global state to be a per-channel state.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Add code to distribute channel interrupt load
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:50 +0000 (06:46 -0800)]
Drivers: hv: Add code to distribute channel interrupt load

Implement a simple policy for distributing incoming interrupt load.
We classify channels as (a) performance critical and (b) not
performance critical. All non-performance critical channels will
be bound to the boot cpu. Performance critical channels will be
bound to the remaining available CPUs on a round-robin basis.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Modify the interrupt handling code to support win8 and beyond
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:49 +0000 (06:46 -0800)]
Drivers: hv: Modify the interrupt handling code to support win8 and beyond

Starting with Win8 (WS2012), the event page can be used to directly get the
channel ID that needs servicing. Modify the channel event handling code
to take advantage of this feature.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Add state to manage incoming channel interrupt load
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:48 +0000 (06:46 -0800)]
Drivers: hv: Add state to manage incoming channel interrupt load

Add state to bind a channel to a specific VCPU. This will help us better
distribute incoming interrupt load.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Setup a mapping for Hyper-V's notion cpu ID
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:47 +0000 (06:46 -0800)]
Drivers: hv: Setup a mapping for Hyper-V's notion cpu ID

On win8 (ws2012), incoming vmbus interrupt load can be spread across all
available VCPUs in the guest. On a per-channel basis, the interrupts can
be bound to specific CPUs. The Linux notion of cpu ID may be different
from that of the hypervisor's. Setup a mapping structure.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Cleanup vmbus_set_event() to support win7 and beyond
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:46 +0000 (06:46 -0800)]
Drivers: hv: Cleanup vmbus_set_event() to support win7 and beyond

On win7 (ws2008 R2) and beyond, we have the notion of having dedicated interrupts on
a per-channel basis. When a channel has a dedicated interrupt assigned, there is no need
to set the interrupt bit in the shared page. Implement this optimization.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Manage signaling state on a per-connection basis
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:45 +0000 (06:46 -0800)]
Drivers: hv: Manage signaling state on a per-connection basis

The current code has a global handle for supporting signaling of the host
from guest. Make this a per-channel attribute as on some versions of the
host we can signal on per-channel handle.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Move vmbus version definitions to hyperv.h
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:44 +0000 (06:46 -0800)]
Drivers: hv: Move vmbus version definitions to hyperv.h

To support version specific optimization in various vmbus drivers,
move the vmbus definitions to the public header file.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Change the signature of vmbus_set_event()
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:43 +0000 (06:46 -0800)]
Drivers: hv: Change the signature of vmbus_set_event()

In preparation for supporting a per-connection signaling mechanism,
change the signature of vmbus_set_event(). This change is also
needed to implement other aspects of the signaling optimization.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Change the signature for hv_signal_event()
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:42 +0000 (06:46 -0800)]
Drivers: hv: Change the signature for hv_signal_event()

In preparation for implementing a per-connection signaling framework,
change the signature of the function hv_signal_event(). The current
code uses a global handle for signaling the host.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Save and export negotiated vmbus version
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:41 +0000 (06:46 -0800)]
Drivers: hv: Save and export negotiated vmbus version

Export the negotiated vmbus version as this may be useful for
individual drivers.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Extend/modify vmbus_channel_offer_channel for win7 and beyond
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:40 +0000 (06:46 -0800)]
Drivers: hv: Extend/modify vmbus_channel_offer_channel for win7 and beyond

The "offfer" message sent by the host has been extended in win7 (ws2008 R2).
Add/modify state to reflect this extension. All these changes are backward
compatible.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Update the ring buffer structure to match win8 functionality
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:39 +0000 (06:46 -0800)]
Drivers: hv: Update the ring buffer structure to match win8 functionality

Update the ringbuffer structure to support win8 functionality.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Support handling multiple VMBUS versions
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:38 +0000 (06:46 -0800)]
Drivers: hv: Support handling multiple VMBUS versions

The current code hard coded the vmbus version independent of the host
it was running on. Add code to dynamically negotiate the most appropriate
version.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Get rid of hv_get_ringbuffer_interrupt_mask()
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:37 +0000 (06:46 -0800)]
Drivers: hv: Get rid of hv_get_ringbuffer_interrupt_mask()

This function is no longer used; get rid of it.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Optimize the signaling on the write path
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:36 +0000 (06:46 -0800)]
Drivers: hv: Optimize the signaling on the write path

The host has already implemented the "read" side optimizations.
Leverage that to optimize "write" side signaling.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Optimize signaling in the read path
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:35 +0000 (06:46 -0800)]
Drivers: hv: Optimize signaling in the read path

Now that we have the infratructure for correctly determining when we
should signal the host; optimize the signaling on the read side -
signaling the guest from the host.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Turn off batched reading for util drivers
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:34 +0000 (06:46 -0800)]
Drivers: hv: Turn off batched reading for util drivers

Util driver is not a performance critical driver and furthermore some
util services such as KVP can only handle one outstanding message
from the host. Turn off batched reading for util drivers.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Add state to manage batched reading
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:33 +0000 (06:46 -0800)]
Drivers: hv: Add state to manage batched reading

For the "read" side signaling optimization, the reader has to completely
drain the queue before exiting. Add state to manage this "batched"
reading.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoDrivers: hv: Implement routines for read side signaling optimization
K. Y. Srinivasan [Sat, 1 Dec 2012 14:46:32 +0000 (06:46 -0800)]
Drivers: hv: Implement routines for read side signaling optimization

Implement functions that will support read-side signaling optimization.
By having the reader indicate the start of the "read" operation and the
"end" of the read operation we can more efficiently handle the signaling
protocol: while the read is in progress, there is no need for the "writer"
to signal the "reader" as new items are put on the read queue.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agotools/hv: Fix permissions of created directory and files
Ben Hutchings [Tue, 27 Nov 2012 07:56:34 +0000 (08:56 +0100)]
tools/hv: Fix permissions of created directory and files

It's silly to create directories without execute permission, or to
give permissions to 'other' but not the group-owner.

Write the permissions in octal and 'ls -l' format since these are much
easier to read than the named macros.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Tomas Hozza <thozza@redhat.com>
Acked-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agotools/hv: Fix /var subdirectory
Tomas Hozza [Tue, 27 Nov 2012 07:56:33 +0000 (08:56 +0100)]
tools/hv: Fix /var subdirectory

Initial patch by Ben Hutchings <ben@decadent.org.uk>

We will install this in /usr, so it must use /var/lib for its state.
Only programs installed under /opt should use /var/opt.

Signed-off-by: Tomas Hozza <thozza@redhat.com>
Acked-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agohv: hv_balloon: remove duplicated include from hv_balloon.c
Wei Yongjun [Thu, 29 Nov 2012 02:22:43 +0000 (21:22 -0500)]
hv: hv_balloon: remove duplicated include from hv_balloon.c

Remove duplicated include.

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Acked-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agomisc: lattice-ecp3-config.c: remove __dev* markings
Greg Kroah-Hartman [Thu, 17 Jan 2013 18:25:26 +0000 (10:25 -0800)]
misc: lattice-ecp3-config.c: remove __dev* markings

These are now removed from the kernel, so remove them to allow the
driver to build properly.

Cc: Stefan Roese <sr@denx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agomisc: Add Lattice ECP3 FPGA configuration via SPI
Stefan Roese [Fri, 7 Dec 2012 08:06:59 +0000 (09:06 +0100)]
misc: Add Lattice ECP3 FPGA configuration via SPI

This patch adds support for bitstream configuration (programming /
loading) of the Lattice ECP3 FPGA's via the SPI bus.

Here an example on my custom MPC5200 based board:

$ echo 1 > /sys/class/firmware/spi0.0/loading
$ cat fpga_a4m2k.bit > /sys/class/firmware/spi0.0/data
$ echo 0 > /sys/class/firmware/spi0.0/loading

leads to these messages:

lattice-ecp3 spi0.0: FPGA Lattice ECP3-35 detected
lattice-ecp3 spi0.0: Configuring the FPGA...
lattice-ecp3 spi0.0: FPGA succesfully configured!

Signed-off-by: Stefan Roese <sr@denx.de>
Acked-by: Ming Lei <ming.lei@canonical.com>
Reviewed-by: Grant Likely <grant.likely@secretlab.ca>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agouio-howto: example bug
Stephen Hemminger [Wed, 16 Jan 2013 18:05:32 +0000 (10:05 -0800)]
uio-howto: example bug

Bug in demo program, checking wrong return value

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Cc: "Hans J. Koch" <hjk@hansjkoch.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agomisc: st_core: Error triggered by convert "char" to "int"
channing [Thu, 10 Jan 2013 08:27:29 +0000 (16:27 +0800)]
misc: st_core: Error triggered by convert "char" to "int"

When st driver decodes protocol index received from raw data,
it does a value convert from "char" to "int". Because it's sign
extension from bit8 to bit32, the "int" value maybe minus, in
another word, the protocol index might be minus, but driver doesn't
filter such case and may continue access memory pointed by this
minus index.

This patch is to change the variable type of index from "int"
to "unsigned char", so that it avoids do such kind of type
conversion.

cc: liu chuansheng <chuansheng.liu@intel.com>
Signed-off-by: channing <chao.bi@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
Linus Torvalds [Thu, 17 Jan 2013 16:56:30 +0000 (08:56 -0800)]
Merge branch 'for-linus' of git://git./linux/kernel/git/s390/linux

Pull more s390 patches from Martin Schwidefsky:
 "A couple of bug fixes: one of the transparent huge page primitives is
  broken, the sched_clock function overflows after 417 days, the XFS
  module has grown too large for -fpic and the new pci code has broken
  normal channel subsystem notifications."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390/chsc: fix SEI usage
  s390/time: fix sched_clock() overflow
  s390: use -fPIC for module compile
  s390/mm: fix pmd_pfn() for thp

11 years agoMerge tag 'for-linus-v3.8-rc4' of git://oss.sgi.com/xfs/xfs
Linus Torvalds [Thu, 17 Jan 2013 00:19:54 +0000 (16:19 -0800)]
Merge tag 'for-linus-v3.8-rc4' of git://oss.sgi.com/xfs/xfs

Pull xfs bugfixes from Ben Myers:

 - fix(es) for compound buffers

 - fix for dquot soft timer asserts due to overflow of d_blk_softlimit

 - fix for regression in dir v2 code introduced in commit 20f7e9f3726a
   ("xfs: factor dir2 block read operations")

* tag 'for-linus-v3.8-rc4' of git://oss.sgi.com/xfs/xfs:
  xfs: recalculate leaf entry pointer after compacting a dir2 block
  xfs: remove int casts from debug dquot soft limit timer asserts
  xfs: fix the multi-segment log buffer format
  xfs: fix segment in xfs_buf_item_format_segment
  xfs: rename bli_format to avoid confusion with bli_formats
  xfs: use b_maps[] for discontiguous buffers

11 years agoMerge tag 'pm+acpi-for-3.8-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 16 Jan 2013 22:34:52 +0000 (14:34 -0800)]
Merge tag 'pm+acpi-for-3.8-rc4' of git://git./linux/kernel/git/rafael/linux-pm

Pull ACPI and power management fixes from Rafael Wysocki:

 - cpuidle regression fix related to the initialization of state
   kobjects from Krzysztof Mazur.

 - cpuidle fix removing some not very useful code and making some
   user-visible problems go away at the same time.  From Daniel Lezcano.

 - ACPI build fix from Yinghai Lu.

* tag 'pm+acpi-for-3.8-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  cpuidle: remove the power_specified field in the driver
  ACPI / glue: Fix build with ACPI_GLUE_DEBUG set
  cpuidle: fix number of initialized/destroyed states

11 years agoxfs: recalculate leaf entry pointer after compacting a dir2 block
Eric Sandeen [Thu, 10 Jan 2013 16:41:48 +0000 (10:41 -0600)]
xfs: recalculate leaf entry pointer after compacting a dir2 block

Dave Jones hit this assert when doing a compile on recent git, with
CONFIG_XFS_DEBUG enabled:

XFS: Assertion failed: (char *)dup - (char *)hdr == be16_to_cpu(*xfs_dir2_data_unused_tag_p(dup)), file: fs/xfs/xfs_dir2_data.c, line: 828

Upon further digging, the tag found by xfs_dir2_data_unused_tag_p(dup)
contained "2" and not the proper offset, and I found that this value was
changed after the memmoves under "Use a stale leaf for our new entry."
in xfs_dir2_block_addname(), i.e.

                        memmove(&blp[mid + 1], &blp[mid],
                                (highstale - mid) * sizeof(*blp));

overwrote it.

What has happened is that the previous call to xfs_dir2_block_compact()
has rearranged things; it changes btp->count as well as the
blp array.  So after we make that call, we must recalculate the
proper pointer to the leaf entries by making another call to
xfs_dir2_block_leaf_p().

Dave provided a metadump image which led to a simple reproducer
(create a particular filename in the affected directory) and this
resolves the testcase as well as the bug on his live system.

Thanks also to dchinner for looking at this one with me.

Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Tested-by: Dave Jones <davej@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Mark Tinguely <tinguely@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
11 years agoxfs: remove int casts from debug dquot soft limit timer asserts
Brian Foster [Fri, 21 Dec 2012 15:45:17 +0000 (10:45 -0500)]
xfs: remove int casts from debug dquot soft limit timer asserts

The int casts here make it easy to trigger an assert with a large
soft limit. For example, set a >4TB soft limit on an empty volume
to reproduce a (0 > -x) comparison due to an overflow of
d_blk_softlimit.

Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Ben Myers <bpm@sgi.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
11 years agoxfs: fix the multi-segment log buffer format
Mark Tinguely [Tue, 4 Dec 2012 23:18:05 +0000 (17:18 -0600)]
xfs: fix the multi-segment log buffer format

Per Dave Chinner suggestion, this patch:
 1) Corrects the detection of whether a multi-segment buffer is
    still tracking data.
 2) Clears all the buffer log formats for a multi-segment buffer.

Signed-off-by: Mark Tinguely <tinguely@sgi.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
11 years agoxfs: fix segment in xfs_buf_item_format_segment
Mark Tinguely [Tue, 4 Dec 2012 23:18:04 +0000 (17:18 -0600)]
xfs: fix segment in xfs_buf_item_format_segment

Not every segment in a multi-segment buffer is dirty in a
transaction and they will not be outputted. The assert in
xfs_buf_item_format_segment() that checks for the at least
one chunk of data in the segment to be used is not necessary
true for multi-segmented buffers.

Signed-off-by: Mark Tinguely <tinguely@sgi.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
11 years agoxfs: rename bli_format to avoid confusion with bli_formats
Mark Tinguely [Tue, 4 Dec 2012 23:18:03 +0000 (17:18 -0600)]
xfs: rename bli_format to avoid confusion with bli_formats

Rename the bli_format structure to __bli_format to avoid
accidently confusing them with the bli_formats pointer.

Signed-off-by: Mark Tinguely <tinguely@sgi.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Ben Myers <bpm@sgi.com>
11 years agoxfs: use b_maps[] for discontiguous buffers
Mark Tinguely [Tue, 4 Dec 2012 23:18:02 +0000 (17:18 -0600)]
xfs: use b_maps[] for discontiguous buffers

Commits starting at 77c1a08 introduced a multiple segment support
to xfs_buf. xfs_trans_buf_item_match() could not find a multi-segment
buffer in the transaction because it was looking at the single segment
block number rather than the multi-segment b_maps[0].bm.bn. This
results on a recursive buffer lock that can never be satisfied.

This patch:
 1) Changed the remaining b_map accesses to be b_maps[0] accesses.
 2) Renames the single segment b_map structure to __b_map to avoid
    future confusion.

Signed-off-by: Mark Tinguely <tinguely@sgi.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Ben Myers <bpm@sgi.com>
11 years agoTell the world we gave up on pushing CC_OPTIMIZE_FOR_SIZE
Kirill Smelkov [Fri, 2 Nov 2012 11:41:01 +0000 (15:41 +0400)]
Tell the world we gave up on pushing CC_OPTIMIZE_FOR_SIZE

In commit 281dc5c5ec0f ("Give up on pushing CC_OPTIMIZE_FOR_SIZE") we
already changed the actual default value, but the help-text still
suggested 'y'. Fix the help text too, for all the same reasons.

Sadly, -Os keeps on generating some very suboptimal code for certain
cases, to the point where any I$ miss upside is swamped by the downside.
The main ones are:

 - using "rep movsb" for memcpy, even on CPU's where that is
   horrendously bad for performance.

 - not honoring branch prediction information, so any I$ footprint you
   win from smaller code, you lose from less code density in the I$.

 - using divide instructions when that is very expensive.

Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomfd, TWL4030: TWL4030 need select REGMAP_I2C
Chuansheng Liu [Mon, 24 Dec 2012 14:19:56 +0000 (22:19 +0800)]
mfd, TWL4030: TWL4030 need select REGMAP_I2C

Fix the build error:

  drivers/built-in.o: In function `twl_probe':
  drivers/mfd/twl-core.c:1256: undefined reference to `devm_regmap_init_i2c'
  make: *** [vmlinux] Error 1

Signed-off-by: liu chuansheng <chuansheng.liu@intel.com>
Acked-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
[ Samuel is busy, taking it directly  - Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agodrivers/base/cpu.c: Fix typo in comment
Ralf Baechle [Tue, 15 Jan 2013 14:27:46 +0000 (15:27 +0100)]
drivers/base/cpu.c: Fix typo in comment

[ We should make fun of people who can't speel too, but then we'd have
  no time for any real work at all  - Linus ]

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agolockdep, rwsem: fix down_write_nest_lock() if !CONFIG_DEBUG_LOCK_ALLOC
Jiri Kosina [Tue, 15 Jan 2013 19:12:37 +0000 (20:12 +0100)]
lockdep, rwsem: fix down_write_nest_lock() if !CONFIG_DEBUG_LOCK_ALLOC

Commit 1b963c81b145 ("lockdep, rwsem: provide down_write_nest_lock()")
contains a bug in a codepath when CONFIG_DEBUG_LOCK_ALLOC is disabled,
which causes down_read() to be called instead of down_write() by mistake
on such configurations.  Fix that.

Reported-and-tested-by: Andrew Clayton <andrew@digital-domain.net>
Reported-and-tested-by: Zlatko Calusic <zlatko.calusic@iskon.hr>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Reviewed-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agoMerge tag 'sound-3.8' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound
Linus Torvalds [Wed, 16 Jan 2013 19:33:52 +0000 (11:33 -0800)]
Merge tag 'sound-3.8' of git://git./linux/kernel/git/tiwai/sound

Pull second round of sound fixes from Takashi Iwai:
 "Yet a few more fixes popped up in this week.

  The biggest change here is the addition of pinctrl support for Atmel,
  which turned out to be almost mandatory to make things working.

  The rest are a few fixes for M-Audio usb-audio device and a fix for
  regression of HD-audio HDMI codecs with alsactl in the recent kernel."

* tag 'sound-3.8' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  ALSA: hda/hdmi - Work around "alsactl restore" errors
  ALSA: usb-audio: selector map for M-Audio FT C400
  ALSA: usb-audio: M-Audio FT C400 skip packet quirk
  ALSA: usb-audio: correct M-Audio C400 clock source quirk
  ALSA: usb - fix race in creation of M-Audio Fast track pro driver
  ASoC: atmel-ssc: add pinctrl selection to driver
  ARM: at91/dts: add pinctrl support for SSC peripheral

11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending
Linus Torvalds [Wed, 16 Jan 2013 19:13:39 +0000 (11:13 -0800)]
Merge git://git./linux/kernel/git/nab/target-pending

Pull scsi target fixes from Nicholas Bellinger:
 "This includes an important >= v3.6 regression bugfix for active I/O
  shutdown (Roland), some TMR related failure / corner cases fixes for
  long outstanding I/O (Roland), two FCoE target mode fabric fabric role
  fixes (MDR), a fix for an incorrect sense code during LUN
  communication failure (Dr. Hannes), plus a handful of other minor
  fixes.

  There are still some outstanding zero-length control CDB regression
  fixes that need to be addressed for v3.8, that will be coming in a
  follow-up PULL request."

* git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending:
  iscsi-target: Fix CmdSN comparison (use cmd->cmd_sn instead of cmd->stat_sn)
  target: Release se_cmd when LUN lookup fails for TMR
  target: Fix use-after-free in LUN RESET handling
  target: Fix missing CMD_T_ACTIVE bit regression for pending WRITEs
  tcm_fc: Do not report target role when target is not defined
  tcm_fc: Do not indicate retry capability to initiators
  target: Use TCM_NO_SENSE for initialisation
  target: Introduce TCM_NO_SENSE
  target: use correct sense code for LUN communication failure

11 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs
Linus Torvalds [Wed, 16 Jan 2013 18:55:10 +0000 (10:55 -0800)]
Merge branch 'for_linus' of git://git./linux/kernel/git/jack/linux-fs

Pull ext3 and udf fixes from Jan Kara:
 "One ext3 performance regression fix and one udf regression fix (oops
  on interrupted mount)."

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs:
  UDF: Fix a null pointer dereference in udf_sb_free_partitions
  jbd: don't wake kjournald unnecessarily

11 years agoMerge git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Wed, 16 Jan 2013 18:17:09 +0000 (10:17 -0800)]
Merge git://git.kernel.org/pub/scm/virt/kvm/kvm

Pull s390 KVM fix from Gleb Natapov.

* git://git.kernel.org/pub/scm/virt/kvm/kvm:
  s390/kvm: Fix BUG in include/linux/kvm_host.h:745

11 years agoMerge tag 'sh-for-linus' of git://github.com/pmundt/linux-sh
Linus Torvalds [Wed, 16 Jan 2013 18:13:04 +0000 (10:13 -0800)]
Merge tag 'sh-for-linus' of git://github.com/pmundt/linux-sh

Pull SuperH fixes from Paul Mundt.

* tag 'sh-for-linus' of git://github.com/pmundt/linux-sh:
  sh: ecovec: add sample amixer settings
  sh: Fix up stack debugging build.
  sh: wire up finit_module syscall.
  sh: Fix FDPIC binary loader
  sh: clkfwk: bugfix: sh_clk_div_enable() care sh_clk_div_set_rate() if div6
  sh: define TASK_UNMAPPED_BASE as a page aligned constant

11 years agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/cmarinas...
Linus Torvalds [Wed, 16 Jan 2013 17:44:40 +0000 (09:44 -0800)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/cmarinas/linux-aarch64

Pull arm64 fixes from Catalin Marinas:
 - Page protection fixes, including proper PAGE_NONE handling
 - Timezone vdso sequence counting fix
 - Additional compat syscall wiring

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/cmarinas/linux-aarch64:
  arm64: compat: add syscall table entries for new syscalls
  arm64: mm: introduce present, faulting entries for PAGE_NONE
  arm64: mm: only wrprotect clean ptes if they are present
  arm64: vdso: remove broken, redundant sequence counting for timezones

11 years agoMerge branch 'x86/urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Linus Torvalds [Wed, 16 Jan 2013 17:11:50 +0000 (09:11 -0800)]
Merge branch 'x86/urgent' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Peter Anvin:
 "This is mainly a workaround for a bug in Sandy Bridge graphics which
  causes corruption of certain memory pages."

* 'x86/urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/Sandy Bridge: Sandy Bridge workaround depends on CONFIG_PCI
  x86/Sandy Bridge: mark arrays in __init functions as __initconst
  x86/Sandy Bridge: reserve pages when integrated graphics is present
  x86, efi: correct precedence of operators in setup_efi_pci

11 years agoMAINTAINERS: update email address for Timur Tabi
Timur Tabi [Tue, 15 Jan 2013 20:19:45 +0000 (14:19 -0600)]
MAINTAINERS: update email address for Timur Tabi

Timur Tabi no longer works for Freescale, so update the email address
and status for all of his maintained projects.

Also mark the QE library as orphaned, for lack of interest in
maintaining it.

The CS4270 driver is marked as "Odd Fixes" because appropriate hardware
is no longer available.

Signed-off-by: Timur Tabi <timur@freescale.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agofirmware: make sure the fw file size is not 0
Luciano Coelho [Tue, 15 Jan 2013 08:43:43 +0000 (10:43 +0200)]
firmware: make sure the fw file size is not 0

If the requested firmware file size is 0 bytes in the filesytem, we
will try to vmalloc(0), which causes a warning:

  vmalloc: allocation failure: 0 bytes
  kworker/1:1: page allocation failure: order:0, mode:0xd2
    __vmalloc_node_range+0x164/0x208
    __vmalloc_node+0x4c/0x58
    vmalloc+0x38/0x44
    _request_firmware_load+0x220/0x6b0
    request_firmware+0x64/0xc8
    wl18xx_setup+0xb4/0x570 [wl18xx]
    wlcore_nvs_cb+0x64/0x9f8 [wlcore]
    request_firmware_work_func+0x94/0x100
    process_one_work+0x1d0/0x750
    worker_thread+0x184/0x4ac
    kthread+0xb4/0xc0

To fix this, check whether the file size is less than or equal to zero
in fw_read_file_contents().

Cc: stable <stable@vger.kernel.org> [3.7]
Signed-off-by: Luciano Coelho <coelho@ti.com>
Acked-by: Ming Lei <ming.lei@canonical.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agomodule, async: async_synchronize_full() on module init iff async is used
Tejun Heo [Wed, 16 Jan 2013 02:52:51 +0000 (18:52 -0800)]
module, async: async_synchronize_full() on module init iff async is used

If the default iosched is built as module, the kernel may deadlock
while trying to load the iosched module on device probe if the probing
was running off async.  This is because async_synchronize_full() at
the end of module init ends up waiting for the async job which
initiated the module loading.

 async A modprobe

 1. finds a device
 2. registers the block device
 3. request_module(default iosched)
4. modprobe in userland
5. load and init module
6. async_synchronize_full()

Async A waits for modprobe to finish in request_module() and modprobe
waits for async A to finish in async_synchronize_full().

Because there's no easy to track dependency once control goes out to
userland, implementing properly nested flushing is difficult.  For
now, make module init perform async_synchronize_full() iff module init
has queued async jobs as suggested by Linus.

This avoids the described deadlock because iosched module doesn't use
async and thus wouldn't invoke async_synchronize_full().  This is
hacky and incomplete.  It will deadlock if async module loading nests;
however, this works around the known problem case and seems to be the
best of bad options.

For more details, please refer to the following thread.

  http://thread.gmane.org/gmane.linux.kernel/1420814

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Alex Riesen <raa.lkml@gmail.com>
Tested-by: Ming Lei <ming.lei@canonical.com>
Tested-by: Alex Riesen <raa.lkml@gmail.com>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agos390/chsc: fix SEI usage
Sebastian Ott [Tue, 15 Jan 2013 18:02:01 +0000 (19:02 +0100)]
s390/chsc: fix SEI usage

cbc0dd1 "s390/pci: CHSC PCI support for error and availability events"
introduced a new SEI notification type as part of pci support.
The way SEI was called with nt2 and nt0 consecutive broke the nt0
stuff used for channel subsystem notifications.

The reason why this was broken with the mentioned patch is that you
cannot selectively disable type 0 notifications (so even when asked
for type 2 only, type 0 could be presented).

The way to do it is to tell SEI which types of notification you can
process and -this is the important part- look at the SEI result which
notification type you actually received.

Reviewed-by: Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
Tested-by: Michael Holzheu <holzheu@linux.vnet.ibm.com>
Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
11 years agos390/time: fix sched_clock() overflow
Heiko Carstens [Mon, 14 Jan 2013 15:55:55 +0000 (16:55 +0100)]
s390/time: fix sched_clock() overflow

Converting a 64 Bit TOD format value to nanoseconds means that the value
must be divided by 4.096. In order to achieve that we multiply with 125
and divide by 512.
When used within sched_clock() this triggers an overflow after appr.
417 days. Resulting in a sched_clock() return value that is much smaller
than previously and therefore may cause all sort of weird things in
subsystems that rely on a monotonic sched_clock() behaviour.

To fix this implement a tod_to_ns() helper function which converts TOD
values without overflow and call this function from both places that
open coded the conversion: sched_clock() and kvm_s390_handle_wait().

Cc: stable@kernel.org
Reviewed-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
11 years agosh: ecovec: add sample amixer settings
Kuninori Morimoto [Tue, 25 Dec 2012 04:04:20 +0000 (20:04 -0800)]
sh: ecovec: add sample amixer settings

FSI - DA7210 needs amixer settings to use it.
This patch adds quick setting guide

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
11 years agow1_therm: Retries: remove old code add CRC
David Stevenson [Tue, 18 Dec 2012 01:37:56 +0000 (01:37 +0000)]
w1_therm: Retries: remove old code add CRC

w1_therm includes some obsolete code to detect bad_roms, this is no
longer relevant.
The retry code is only used for this bad_rom test, however there is a
CRC check that detects a bad read, but does not trigger a retry. This
patch removes all the bad_rom code and uses the CRC check to trigger
retries.

Signed-off-by: David Stevenson <david@avoncliff.com>
Acked-by: Evgeniy Polyakov <zbr@ioremap.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agodrivers/w1/masters/mxc_w1.c: use devm_ functions
Julia Lawall [Thu, 6 Dec 2012 23:15:24 +0000 (00:15 +0100)]
drivers/w1/masters/mxc_w1.c: use devm_ functions

The various devm_ functions allocate memory that is released when a driver
detaches.  This patch uses these functions for data that is allocated in
the probe function of a platform device and is only freed in the remove
function.

At the same time, this fixes two faults.  First, mdev, the result of
kzalloc, was never freed.  Second, on failure of ioremap, 0 was returned.
This has been replaced by -EBUSY, which was the failure value for the call
to request_mem_region, with which the call to ioremap has been combined.

The warning message on failure of ioremap is dropped, because
devm_request_and_ioremap already gives such messages on failure.

Finally, the initial call to platform_get_resource is moved closer to the
call to devm_request_and_ioremap, which takes care of checking whether its
result is NULL, implying that a test on the result of this call to
platform_get_resource is not needed.

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Acked-by: Evgeniy Polyakov <zbr@ioremap.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agodrivers/w1/masters/ds1wm.c: use devm_ functions
Julia Lawall [Thu, 6 Dec 2012 23:15:23 +0000 (00:15 +0100)]
drivers/w1/masters/ds1wm.c: use devm_ functions

The various devm_ functions allocate memory that is released when a driver
detaches.  This patch uses these functions for data that is allocated in
the probe function of a platform device and is only freed in the remove
function.

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Acked-by: Evgeniy Polyakov <zbr@ioremap.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoMerge tag 'extcon-for-3.9' of git://git.kernel.org/pub/scm/linux/kernel/git/chanwoo...
Greg Kroah-Hartman [Wed, 16 Jan 2013 07:26:35 +0000 (23:26 -0800)]
Merge tag 'extcon-for-3.9' of git://git./linux/kernel/git/chanwoo/extcon into char-misc-next

Chanwoo writes:
  This is patch set of extcon for v3.9.

  Update to max77693/max8997 extcon driver:
  - Using MHL_TA cable for charging.
  - Support JIG cable.
  - Support Dock-Audio device for playing music and button of device.
  - Support Dock-Smart device for desktop mode with mouse/keyboard.
  - Set default UART/USB path on probe().
  - Check the state/type of cable after completing initialization.
  - Code clean to remove duplicate code and bug fix related to sequence of interrupt.
  - Fix irq_flag of max8997/max77693 driver.

  Update to arizona extcon driver:
  - Headphone measurements.
  - Alternative detection mechanism for non-default system designs.
  - Microphone clamp integration.
  - Support for additional detection pin.
  - MICBIAS rise time configuration.

11 years agoipack/devices/ipoctal: added shutdown callback
Samuel Iglesias Gonsalvez [Mon, 10 Dec 2012 10:50:08 +0000 (11:50 +0100)]
ipack/devices/ipoctal: added shutdown callback

Added shutdown callback to disable RX and TX when there is no other client
accesing the device.

Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoipack/devices/ipoctal: add rx_enable flag
Samuel Iglesias Gonsalvez [Mon, 10 Dec 2012 10:50:07 +0000 (11:50 +0100)]
ipack/devices/ipoctal: add rx_enable flag

Thus, we don't enable RX when a termios setup has been called, as it could be
disabled previously.

As the control registers (Rx, Tx flags specifically) cannot be read from the
device, we keep this info in rx_enable.

Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoipack/devices/ipoctal: remove redundant tty_flip_buffer_push()
Samuel Iglesias Gonsalvez [Mon, 10 Dec 2012 10:50:06 +0000 (11:50 +0100)]
ipack/devices/ipoctal: remove redundant tty_flip_buffer_push()

The function is already called in ipoctal_irq_rx()

Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoipack/devices/ipoctal: protect the channel data processing with a spinlock
Samuel Iglesias Gonsalvez [Mon, 10 Dec 2012 10:50:05 +0000 (11:50 +0100)]
ipack/devices/ipoctal: protect the channel data processing with a spinlock

We protect important data such as TX buffer pointer, nb_bytes counter and status
registers of the device, from accessing several times at the same time.

Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoipack/devices/ipoctal: ack IRQ before processing it
Samuel Iglesias Gonsalvez [Mon, 10 Dec 2012 10:50:04 +0000 (11:50 +0100)]
ipack/devices/ipoctal: ack IRQ before processing it

Due to the IRQ processing, we can generate another IRQ that can come before we
end the previous one, so we lost it. E.g. when transmitting a character.

To allow the processing in SMP machines, we ack the IRQ at the beginning of the
IRQ handler.

Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoipack/devices/ipoctal: avoid re-enable RX two times.
Samuel Iglesias Gonsalvez [Mon, 10 Dec 2012 10:50:03 +0000 (11:50 +0100)]
ipack/devices/ipoctal: avoid re-enable RX two times.

RX is enabled when the tty port is open, so no need to do it in initialization
time: it can allow the device to receive characters but no TTY client is
listening to them.

It produced an infinite number of IRQ as RxFIFO is not read to clear that
IRQ in the device, so it is still pending.

Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoipack/devices/ipoctal: rework disable TX when the TX buffer is empty
Samuel Iglesias Gonsalvez [Mon, 10 Dec 2012 10:50:02 +0000 (11:50 +0100)]
ipack/devices/ipoctal: rework disable TX when the TX buffer is empty

Depending of the device, it disables the TX mode in different places when there
is no more data to transmit.

This patch reorder them and disable the TX mode in the same place.

Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoipack/devices/ipoctal: setup TTY_NORMAL flag for each character.
Samuel Iglesias Gonsalvez [Mon, 10 Dec 2012 10:50:01 +0000 (11:50 +0100)]
ipack/devices/ipoctal: setup TTY_NORMAL flag for each character.

In case of several characters present in RxFIFO, they will have the flag of the
previous one, no matter if the actual character was received properly or not.

This patch fixes this bug.

Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoipack/devices/ipoctal: remove wait_queue and atomic_t board_write
Samuel Iglesias Gonsalvez [Mon, 10 Dec 2012 10:50:00 +0000 (11:50 +0100)]
ipack/devices/ipoctal: remove wait_queue and atomic_t board_write

Don't block the TTY client when sending characters.

Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoipack/devices/ipoctal: fix kernel bug when using pppd
Samuel Iglesias Gonsalvez [Mon, 10 Dec 2012 10:49:59 +0000 (11:49 +0100)]
ipack/devices/ipoctal: fix kernel bug when using pppd

Trying to setup the pppd server to use ipoctal's serial ports, it says the ports
are busy the first time. If the operation is repeated, a kernel bug due to a
dereference of a NULL pointer appears.

Removing the one-access-only setup from the driver, removes this kernel bug.

Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoipack/devices/ipoctal: don't check if nb_bytes is < 0
Alberto Garcia [Mon, 10 Dec 2012 10:49:58 +0000 (11:49 +0100)]
ipack/devices/ipoctal: don't check if nb_bytes is < 0

It is an unsigned int so that check is pointless.

Signed-off-by: Alberto Garcia <agarcia@igalia.com>
Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoipack/devices/ipoctal: Fix race condition during Tx
Alberto Garcia [Mon, 10 Dec 2012 10:49:57 +0000 (11:49 +0100)]
ipack/devices/ipoctal: Fix race condition during Tx

In order to transmit data, the driver enables Tx and sleeps until
*board_write is set to 1 by the interrupt handler.

It can happen, though, that the data is sent even before the process
is asleep. In this case *board_write must be set to 1 anyway,
otherwise we will be waiting for a condition that will never be true.

Signed-off-by: Alberto Garcia <agarcia@igalia.com>
Signed-off-by: Samuel Iglesias Gonsalvez <siglesias@igalia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
11 years agoarm64: compat: add syscall table entries for new syscalls
Will Deacon [Mon, 14 Jan 2013 14:45:46 +0000 (14:45 +0000)]
arm64: compat: add syscall table entries for new syscalls

There have been a number of new syscalls introduced to arch/arm/ since
the compat layer was implemented for arm64, so add pointers to the
relevant functions to the compat syscall table.

Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
11 years agoALSA: hda/hdmi - Work around "alsactl restore" errors
Takashi Iwai [Tue, 15 Jan 2013 13:44:41 +0000 (14:44 +0100)]
ALSA: hda/hdmi - Work around "alsactl restore" errors

When "alsactl restore" is performed on HDMI codecs, it tries to
restore the channel map value since the channel map controls are
writable.  But hdmi_chmap_ctl_put() returns -EBADFD when no PCM stream
is assigned yet, and this results in an error message from alsactl.
Although the error is harmless, it's certainly ugly and can be
regarded as a regression.

As a workaround, this patch changes the return code in such a case to
be zero for making others happy.  (A slight excuse is: when the chmap
is changed through the proper alsa-lib API, the PCM status is checked
there anyway, so we don't have to be too strict in the kernel side.)

Cc: <stable@vger.kernel.org> [v3.7+]
Signed-off-by: Takashi Iwai <tiwai@suse.de>